Pore-C tutorial: haplotype-aware scaffolding from hifiasm hap assemblies (tetraploid example)¶

This tutorial shows how to construct pre-separated haplotypes of a tetraploid genome using Hifiasm (with Pore-C converted Pseudo-Hi-C data), and then scaffold them using C-Phasing in haplotype-aware mode (--mode hapaware).

The goal is to keep haplotypes consistent while producing chromosome-scale scaffolds.

1. Haplotype-Resolved Contig Assembly (Pore-C to Pseudo-Hi-C)¶

Before scaffolding, we generate haplotype-resolved contigs by converting Pore-C reads into virtual paired-end "pseudo-Hi-C" reads to guide Hifiasm's phasing.

Step 1.1: Generate Initial Unitigs¶

First, run hifiasm in default mode using PacBio HiFi reads to generate the primary unitigs, and convert the GFA graph structure to FASTA format:

~/software/hifiasm-0.25.0/hifiasm -t 100 \
    -o hifi.asm \
    hifi.fastq.gz

gfatools gfa2fa hifi.asm.p_utg.gfa > hifi.asm.p_utg.fasta

Step 1.2: Map Pore-C Reads to Unitigs¶

Map your Pore-C reads back to the generated unitigs to capture multi-way contact information:

cphasing mapper hifi.asm.p_utg.fasta porec_reads.fastq.gz -t 100 -o porec_reads.porec.gz

Step 1.3: Convert Pore-C to Pseudo-Hi-C Reads¶

Convert the multi-contact Pore-C alignments into virtual paired-end "pseudo-Hi-C" reads (porec2hic_R1.fa.gz and porec2hic_R2.fa.gz):

cphasing-rs porec2 porec_reads.porec.gz -l 0 -o porec2hic

Step 1.4: Resolve Assembly with Hifiasm (Hi-C Mode)¶

Run hifiasm in Hi-C mode using the PacBio HiFi reads along with the newly generated pseudo-Hi-C paired-end reads to obtain phased contig sets:

Note

Executing this command in the same directory as Step 1.1 will reuse the cached files, allowing hifiasm to bypass the time-consuming HiFi read correction and overlapping phases.

~/software/hifiasm-0.25.0/hifiasm -t 100 \
    -o hifi.asm \
    hifi.fastq.gz \
    --h1 porec2hic_R1.fa.gz \
    --h2 porec2hic_R2.fa.gz

This will produce phased assemblies for haplotypes (e.g., hifi.asm.hic.hap1.p_ctg.fasta to hifi.asm.hic.hap4.p_ctg.fasta).

2. Prepare a combined FASTA¶

C-Phasing expects a single assembly FASTA as the contig set to scaffold.
For hap-aware scaffolding, concatenate your output haplotype assemblies into a single FASTA file.

cat hifi.asm.hic.hap1.p_ctg.fasta \
    hifi.asm.hic.hap2.p_ctg.fasta \
    hifi.asm.hic.hap3.p_ctg.fasta \
    hifi.asm.hic.hap4.p_ctg.fasta > haps.concat.fa

Important: unique contig names¶

Ensure contig IDs are unique across haplotypes. If hifiasm already prefixes contig names per hap (which it does by default), you are fine. Otherwise, rename contigs to avoid collisions.

3. Provide an initial haplotype clustering file (`-fc`)¶

--mode hapaware is designed for cases where haplotypes are already separated.
Providing an initial cluster file helps C-Phasing keep haplotypes consistent.

cphasing utils fa2cluster haps.concat.fa -o haps.clusters.txt

Create a haps.clusters.txt with two columns:

group_id
contig_id

Example (illustrative):

h1   h1tg000001l
h1   h1tg000002l
h2   h2tg000001l
h2   h2tg000002l
h3   h3tg000001l
h4   h4tg000001l

Tip: In hap-aware mode, -fc haps.clusters.txt is the key input that encodes haplotype priors.

4. Run C-Phasing (Pore-C + hapaware)¶

Run the main pipeline with original Pore-C reads and --mode hapaware. Here we scaffold a tetraploid with a basic chromosome number of 12 ($2n = 4x = 48$, so -n 4:12):

cphasing pipeline \
  -f haps.concat.fa \
  -pcd porec_reads.fastq.gz \
  --mode hapaware \
  -fc haps.clusters.txt \
  -o cphasing_hapaware_out \
  -t 32 -hcr -p AAGCTT -n 4:12